Chinese AI start-up DeepSeek breaks for Lunar New Year as its success rattles Wall Street
The Hangzhou office of China’s hottest start-up has recently seen several uninvited guests trying to see the team responsible for shaking up AI
Yet the Hangzhou-based start-up, including founder Liang Wenfeng and the firm’s young scientists, has shunned public attention as China entered its week-long Lunar New Year holiday. The company made its last update at midnight on Monday – the day before Lunar New Year’s Eve, a traditional festival for family reunions – with the launch of its first multimodal model, Janus-Pro. The 7 billion-parameter version of the image generation model outperformed OpenAI’s DALL-E 3 and Stability AI’s Stable Diffusion in benchmark tests, according to the company’s technical report.
During a Tuesday morning visit to its headquarters in Hangzhou, capital of eastern Zhejiang province, the office building where DeepSeek occupies one floor was deserted. A security guard confirmed that no one had been at the office for the day because of the public holiday, but added that there had been many uninvited visitors in the past two days. They were all turned away, as some tried knocking at the door of what has become the hottest tech start-up in the country.
Unlike other tech start-ups, which are often set up at tech parks, the high-rise that houses DeepSeek mainly hosts tenants from the finance industry. The listed address for High-Flyer Quant, the hedge fund owned by DeepSeek’s Liang, is in the same building.
One source who knows the firm told the Post that the company is so low profile that it does not have anyone handling public relations. Another person who is close to the firm said many of the company’s young employees are amazed to see how the world is responding to its cheap-but-high-performing AI models.
